Comparing CFU vs Counts for detecting C. difficile



None

Clindamycin

Vancomycin

Ciprofloxacin

Ampicillin

Cefaperazone

Metronidazole

Streptomycin

diversity vs colonization

alpha diversity

  • Do communities colonized to higher levels have lower diversity (alpha)?
  • Is there an association between cfu and alpha (invsimpson and shannon - NS)?
  • Is there a correlation between cfu and # of otus (NS)?
  • Are there shared otus?

beta diversity

  • Are highly infected communities are most different than untreated?
  • Is there separation between untreated mice and all the highly infected communities (>1e6)?
  • Are communities that recover/elimnate cdifficile are more diverse?
  • Is there difference in diversity between highly infected communities?
  • Is there more change w/low diversity? or more change with high cfu?

  • Need to keep in mind assumption of independent samples when using daily samples?

Alpha Diversity

Communities colonized to a lower level at the end of 10 days have recovered more from the initial antibiotic pertubation

When looking at the alpha diversity of the infected vs uninfected, the community doesnt seem to be different, these communities seem to follow a similar trend regardless of infection. But when looking at the colonization level within the infected, it appears to be a bimodal distributution.

To investigate whether the bimodal distribution of the colonization level vs alpha diversity, we can look at how the communities are distributed prior to infection. If this difference in distribution is related to the level of colonization, we would expect the distribution of end point colonization levels to be random at the time points prior to the infections. When we color the point by end point colonization, it appears to be lower diversity in only the highly colonized mice

So comparing the mice infected with C. difficile, is there a decrease in diversity in the mice that become highly colonized at the final time point? ## Do stats!

## 
##  Wilcoxon signed rank test with continuity correction
## 
## data:  value
## V = 1275, p-value = 7.79e-10
## alternative hypothesis: true location is not equal to 0

Are differences in endpoint diversity not due to colonization but actually abx?

Since the high and low colonization seem to be grouped by antibiotic, is the difference due to antibiotic or cdiff?

How do specific communities transition from abx to infection?

## [[1]]

## [[1]]
## 
## Call:
## lm(formula = get(variable_name) ~ CFU, data = data_frame)
## 
## Residuals:
##    Min     1Q Median     3Q    Max 
## -4.162 -2.412 -1.213  1.429 13.075 
## 
## Coefficients:
##               Estimate Std. Error t value Pr(>|t|)    
## (Intercept)  5.241e+00  1.714e-01   30.58  < 2e-16 ***
## CFU         -1.455e-08  2.479e-09   -5.87 8.24e-09 ***
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 3.272 on 468 degrees of freedom
##   (36 observations deleted due to missingness)
## Multiple R-squared:  0.06859,    Adjusted R-squared:  0.0666 
## F-statistic: 34.46 on 1 and 468 DF,  p-value: 8.244e-09
## [[1]]

## [[1]]
## 
## Call:
## lm(formula = get(variable_name) ~ CFU, data = data_frame)
## 
## Residuals:
##     Min      1Q  Median      3Q     Max 
## -35.753 -15.274  -4.381  12.851  72.405 
## 
## Coefficients:
##               Estimate Std. Error t value Pr(>|t|)    
## (Intercept)  4.609e+01  1.047e+00   44.03  < 2e-16 ***
## CFU         -1.104e-07  1.514e-08   -7.29 1.33e-12 ***
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 19.99 on 468 degrees of freedom
##   (36 observations deleted due to missingness)
## Multiple R-squared:  0.102,  Adjusted R-squared:  0.1001 
## F-statistic: 53.14 on 1 and 468 DF,  p-value: 1.331e-12
## [[1]]

## [[1]]
## 
## Call:
## lm(formula = get(variable_name) ~ CFU, data = data_frame)
## 
## Residuals:
##      Min       1Q   Median       3Q      Max 
## -1.65813 -0.51173 -0.06297  0.45801  2.09371 
## 
## Coefficients:
##               Estimate Std. Error t value Pr(>|t|)    
## (Intercept)  1.899e+00  3.404e-02  55.801  < 2e-16 ***
## CFU         -3.098e-09  4.923e-10  -6.293 7.15e-10 ***
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 0.6498 on 468 degrees of freedom
##   (36 observations deleted due to missingness)
## Multiple R-squared:  0.07803,    Adjusted R-squared:  0.07606 
## F-statistic: 39.61 on 1 and 468 DF,  p-value: 7.149e-10
## [[1]]

## [[1]]
## 
## Call:
## lm(formula = get(variable_name) ~ CFU, data = data_frame)
## 
## Residuals:
##     Min      1Q  Median      3Q     Max 
## -2.9720 -1.5949 -0.8509  0.8313 14.2621 
## 
## Coefficients:
##               Estimate Std. Error t value Pr(>|t|)    
## (Intercept)  4.051e+00  1.503e-01  26.944  < 2e-16 ***
## CFU         -6.395e-09  1.924e-09  -3.324 0.000978 ***
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 2.435 on 366 degrees of freedom
##   (36 observations deleted due to missingness)
## Multiple R-squared:  0.0293, Adjusted R-squared:  0.02665 
## F-statistic: 11.05 on 1 and 366 DF,  p-value: 0.0009777
## [[1]]

## [[1]]
## 
## Call:
## lm(formula = get(variable_name) ~ CFU, data = data_frame)
## 
## Residuals:
##     Min      1Q  Median      3Q     Max 
## -28.215 -12.009  -2.285   9.603  58.318 
## 
## Coefficients:
##               Estimate Std. Error t value Pr(>|t|)    
## (Intercept)  3.882e+01  9.625e-01  40.329  < 2e-16 ***
## CFU         -6.051e-08  1.232e-08  -4.913 1.36e-06 ***
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 15.59 on 366 degrees of freedom
##   (36 observations deleted due to missingness)
## Multiple R-squared:  0.06187,    Adjusted R-squared:  0.0593 
## F-statistic: 24.14 on 1 and 366 DF,  p-value: 1.356e-06
## [[1]]

## [[1]]
## 
## Call:
## lm(formula = get(variable_name) ~ CFU, data = data_frame)
## 
## Residuals:
##      Min       1Q   Median       3Q      Max 
## -1.43246 -0.41705 -0.08106  0.43371  1.47089 
## 
## Coefficients:
##               Estimate Std. Error t value Pr(>|t|)    
## (Intercept)  1.674e+00  3.495e-02  47.885  < 2e-16 ***
## CFU         -1.551e-09  4.473e-10  -3.468 0.000588 ***
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 0.5662 on 366 degrees of freedom
##   (36 observations deleted due to missingness)
## Multiple R-squared:  0.03181,    Adjusted R-squared:  0.02916 
## F-statistic: 12.02 on 1 and 366 DF,  p-value: 0.0005875

After talking with Pat (1/17/18) - Does end point look like pre_abx? - are most similar the recovered ones? - Split analysis by abx/dose/days recovered - Resistance more similar than colonized? - Start with high dose and 1 day recovery - then look at how modulating the dose/recovery affects - train model with low recovery and test with high recovery - Show different context of day 0 - compare differences in metro recovery - how do susceptibility break points compare?

## # A tibble: 8 x 6
##   otu       median_abundance    rho   pvalue pvalue_BH pvalue_bon
##   <chr>                <dbl>  <dbl>    <dbl>     <dbl>      <dbl>
## 1 Otu000003           1.26   -0.614 4.17e-50  7.76e-48   7.76e-48
## 2 Otu000017           0.1    -0.470 3.47e-27  8.35e-26   6.46e-25
## 3 Otu000020           0.250  -0.425 4.71e-22  6.75e-21   8.77e-20
## 4 Otu000004           1.30   -0.408 2.60e-20  3.02e-19   4.83e-18
## 5 Otu000012           0.0513 -0.338 4.62e-14  2.60e-13   8.59e-12
## 6 Otu000011           2.55    0.208 5.26e- 6  1.10e- 5   9.79e- 4
## 7 Otu000015           0.311   0.293 9.04e-11  3.50e-10   1.68e- 8
## 8 Otu000010           1.65    0.482 1.05e-28  3.26e-27   1.96e-26